Geologic Methane Seeps Along Boundaries of Arctic Permafrost Thaw and Melting Glaciers

ARTICLES PUBLISHED ONLINE: 20 MAY 2012 | DOI: 10.1038/NGEO1480

Geologic methane seeps along boundaries of Arctic permafrost thaw and melting glaciers

Katey M. Walter Anthony1*, Peter Anthony1, Guido Grosse2 and Jeffrey Chanton3

Methane, a potent greenhouse gas, accumulates in subsurface hydrocarbon reservoirs, such as coal beds and natural gas deposits. In the Arctic, permafrost and glaciers form a ‘cryosphere cap’ that traps gas leaking from these reservoirs, restricting flow to the atmosphere. With a carbon store of over 1,200 Pg, the Arctic geologic methane reservoir is large when compared with the global atmospheric methane pool of around 5 Pg. As such, the Earth’s climate is sensitive to the escape of even a small fraction of this methane. Here, we document the release of 14C-depleted methane to the atmosphere from abundant gas seeps concentrated along boundaries of permafrost thaw and receding glaciers in Alaska and Greenland, using aerial and ground surface survey data and in situ measurements of methane isotopes and flux. We mapped over 150,000 seeps, which we identified as bubble-induced open holes in lake ice. These seeps were characterized by anomalously high methane fluxes, and in Alaska by ancient radiocarbon ages and stable isotope values that matched those of coal bed and thermogenic methane accumulations. Younger seeps in Greenland were associated with zones of ice-sheet retreat since the Little Ice Age. Our findings imply that in a warming climate, disintegration of permafrost, glaciers and parts of the polar ice sheets could facilitate the transient expulsion of 14C-depleted methane trapped by the cryosphere cap.

ermafrost and glaciers cap large reservoirs of geologic methane Superﬁcial and subcap ebullition seeps 1–5 (CH4) in coal beds, natural gas deposits and as hydrate . The Ebullition (bubbling) is often the dominant pathway of methane Pmechanisms, timescales and rates of exchange of this fossil release from aquatic ecosystems17,18. Here we define two types of (14C-dead) pool with the atmosphere are not well determined. Most methane ebullition seep: superficial and subcap. Superficial seepage of this deeply buried methane reservoir, formed by thermal and refers to the continuous formation and release of ‘ecosystem’ microbial decomposition of organics in sedimentary basins, mi- methane that is formed over relatively recent timescales without grates from source rocks and accumulates under stratigraphic and storage for geologic time. Examples of superficial methane systems structural traps6,7 or, given sufficiently high pressure and low tem- include shallow lake and wetland sediments where methane perature, in ice-like solids in the subsurface known as gas hydrate3. is produced by microbes through anaerobic decomposition of Several authors have advanced the hypothesis that perennially relatively modern organic matter. A second example would be frozen, ice-saturated ground (permafrost) and massive glacial thawed zones under thermokarst lakes formed by degradation overburden form a further impermeable trap for geologic methane of icy, organic-rich yedoma permafrost, where the radiocarbon originating from fossil hydrocarbon reservoirs4,5,8–11 and other non- age of microbial methane reflects the age of recently thawed fossil, but 14C-depleted Late Quaternary deposits buried beneath Pleistocene organic matter17,19. the cryosphere12. Hydrocarbon reservoir models indicate that In contrast, subcap seeps release 14C-depleted methane that permafrost and glacial overburden serve as such a cryosphere has been trapped or impeded by the cryosphere cap. Subcap cap that allows for gas escape during interglacial periods11,13,14. methane may originate from microbial, thermogenic or mixed Gas accumulations have been observed within and beneath microbial–thermogenic processes within sedimentary basins, in- permafrost on the Alaska North Slope and elsewhere in the cluding buried organics associated with glacial sequences, coal Arctic through seismic surveys and permafrost drilling3,15. Despite beds, conventional natural gas reservoirs and potentially methane these hypotheses, model predictions and direct observations of hydrates. Emissions from superficial seeps can be scaled on the basis the cryosphere cap, there is a dearth of field evidence for of existing ecological methods17,20. Quantification of subcap seepage an increase in natural geologic methane seepage as a direct is more challenging because methane accumulations are distributed result of cryosphere disintegration16. This study presents the first beneath complex, site-specific geologic and cryospheric settings. evidence of widespread geologic methane seepage along boundaries Seasonal ice cover on water bodies in the Arctic provides a of cryosphere retreat. unique opportunity to more accurately assess methane ebullition We combined a new method of aerial survey and ground truth from both seep types. Relatively slow bubbling from superficial to identify seep-induced melt-holes in ice-covered water bodies to: seeps produces distinct in-ice bubble patterns and occasionally quantify methane seeps along a north–south transect in Alaska; small holes (0.01–0.3 m2) that remain ice-free for several days confirm the occurrence of anomalous seeps in Greenland; and to weeks following freeze-up17 (Fig. 1a). In contrast, convection document for the first time the widespread occurrence of 14C- associated with anomalously high bubbling rates in subcap seeps depleted methane seeps along boundaries of permafrost thaw and maintains large (up to 300 m2) open holes in ice 0.2–2 m thick melting glaciers in the terrestrial Arctic. that we detected during aerial surveys in winter (Fig. 1b–d). We

1Water and Environmental Research Center, University of Alaska Fairbanks, Alaska 99775-5960, USA, 2Geophysical Institute, University of Alaska Fairbanks, Alaska 99775-7320, USA, 3Department of Earth, Ocean and Atmospheric Science, Florida State University, Florida 32306-4320, USA. *e-mail: [email protected].

ARTICLES NATURE GEOSCIENCE DOI: 10.1038/NGEO1480

1 m 1.5 m

4 m

3 m 25 m

Figure 1 | The effect on lake ice formation of subcap and superficial seeps. a–c, Photographs showing examples of the largest superficial seeps (a) and small (b) and large (c) subcap macroseeps. Even the strongest superficial seeps are ice-covered in late winter. Further, ebullition does not occur simultaneously among superficial seeps (a). In contrast, bubbles breaking the surface of all open holes indicate high, simultaneous ebullition among subcap seeps (b). d, Clustering of subcap seeps is apparent in the aerial photograph. Photographs were taken near Fairbanks, interior Alaska (a), Cook Inlet, southcentral Alaska (b) and Atqasuk, northern Alaska (c,d) one, eight and three weeks, respectively, following freeze-up. Locations are shown in Fig. 2. used aerial and ground surveys combined with ebullition flux and based our analysis on ebullition macroseeps that were visually measurements, gas collection and isotope analyses to map the detectable during winter surveys. Survey results increased the occurrence of superficial and subcap methane seeps along a north– number of previously documented geologic seep-site occurrences south transect in Alaska (Fig. 2). Seeps in West Greenland lakes were in Alaska fourfold (Supplementary Information). quantified only by ground survey (Supplementary Information). We observed most of these cryosphere-cap seeps in lakes Subcap seeps were distinguished from superficial seeps by along the boundaries of permafrost thaw and in moraines exceedingly high ebullition rates (up to 141,600 l seep−1 d−1, and fjords of retreating glaciers. In some regions of Alaska, Fig. 3a), isotopic and geochemical signatures that frequently their discovery increased previous ecosystem-based lake methane matched those of locally identified coal bed methane and natural emission estimates20 by 80–350% (Fig. 4). Ninety per cent of Alaska gas (Fig. 3b–d) and spatial clustering (Fig. 1). Pockmarks up to 3 m subcap seep sites occurred in sedimentary basins, but only 33% diameter in bottom sediments frequently characterized subcap seep overlapped regions of known coal and conventional natural gas sites. Geologic seeps of similar size are known to occur on the sea basins (Fig. 2b). Aerial seep surveys seem to be an excellent tool for floor21,22. However, much of marine seep methane dissolves out hydrocarbon exploration. of bubbles during their ascent through deep water and is lost to microbial oxidation in the ocean’s water column23. In contrast, Subcap seepage due to cryosphere degradation subcap seepage through relatively shallow lakes, rivers and fjord Hydrocarbon seepage worldwide is a function of the occurrence margins escaped directly to the atmosphere with bubble methane of pressurized fluid reservoirs and permeability of the overly- concentrations as high as 99.5% by volume. ing rocks6,8,10. Outside the ice-rich cryosphere, natural gas and coal bed methane escaping from primary structural and strati- Winter ice surveys reveal Alaska subcap seeps graphic traps migrates to the surface through permeable strata, Examination of ∼6,700 lakes across Alaska revealed the occurrence open joints and activated faults and fissures8,21,24–26. In Alaska of 77 previously undocumented subcap seep sites containing and across the pan-Arctic, researchers have demonstrated that >150,000 highly ebullient macroseep vents, which we define as massive glacial ice and permafrost with ice-filled pore space single bubble streams capable of maintaining open holes in ice serve as a further impermeable confining layer that restricts (Fig. 2). We conservatively treated each lake or region of a river gas flow and impedes, slows or focuses gas migration to the containing one or more seep fields as a single subcap-seep site, surface3,4,8,9,11,15. Models predict that the disintegration of the

NATURE GEOSCIENCE DOI: 10.1038/NGEO1480 ARTICLES

a a 60 50 Superficial Atqasuk 40 Subcap 30 20 10

Number of seeps 0 10¬3 10¬2 10¬1 100 101 102 103 104 105 106 Ebullition (l d¬1)

b 120 100 Fairbanks 80 60 40 pmC (%) 20 0 0152025305 10 Lake no. Icy Bay Cook Inlet c ¬90 Katalla Microbial CO2 reduction N ¬70

300 km () Microbial 4 acetate fermentation Early mature Mix CH ¬50 Thermogenic C 13 b δ ¬30 Geothermal crystalline Humic

¬450 ¬400 ¬350 ¬300 ¬250 ¬200 ¬150 ¬100 δ DCH4()

6 d 10 Microbial 105 Acetate CO2 reduction ) fermentation 3 104 Migration Increasing CO2

+ C conversion Increasing 2 103 maturity Oxidation Migration /(C 2 1 10

C Kerogen type II 101 Kerogen type III Thermogenic 100 ¬80 ¬70 ¬60 ¬50 ¬40 ¬30 ¬20 δ 13 CCH4()

Figure 3 | Distinctions between seep types based on bubbling rates and isotope compositions. a–c, Alaska subcap ebullition seeps (filled symbols) were distinct from superficial seeps (open circles and bars) on the basis of vigorous rates of bubbling (a), predominately fossil radiocarbon ages (percentage of modern carbon (pmC), mean±s.e.m. of 1–12 seeps per lake) (b) and enriched stable isotope values originating from thermogenic Seep ebullition (kg CH site¬1 d¬1) 4 and microbial hydrocarbon reservoirs or buried glacial organics (c). Superficial Subcap (macroseep) Lakes 13 0¬1 Yedoma-type deposit d, δ CCH4 and C1/(C2 +C3) ratios in subcap seeps (filled circles) were 1¬10 Coal similar to those in nearby gas wells (open squares) in southcentral Alaska 10¬100 Petroleum (Cook Inlet, blue)36 and northern Alaska near the Meade River33 (green) 100¬1,000 Northern Alaska, continuous and Walakpa33 (orange). Plots c and d are after refs 46 and 47. permafrost (Region 1) 1,000¬10,000 Southcentral Alaska, previously 11,13,14 Previously documented glaciated (Region 2) of glacial ice overburden . Similar predictions were recently geologic gas seeps 49 Interior Alaska, discontinuous to made for CO2 emissions from deep reservoirs . We used geospa- Aerial survey flight path isolated permafrost (Region 3) tial analysis (Supplementary Methods) to test four hypotheses Hydrate accumulations Last Glacial Maximum glaciation Mud volcanoes and (H1–H4) about geologic methane release specific to three ge- hydrothermal gas seeps Modern glaciers ographic regions in Alaska, distinct according to the character of the cryosphere cap. Figure 2 | Alaska map of surveyed methane seeps. a, Yellow dots, Region 1 (northern Alaska, Fig. 2b) represents a type of representing 77 subcap seep sites identified across Alaska in this study, cryosphere cap where geologic methane reservoirs are sealed long- and green dots (superficial study lakes) are scaled by the magnitude of term by thick continuous permafrost. Warming of permafrost −1 −1 methane flux (kg CH4 site d ) at each site. Black dashed lines show and melting of ground ice during the formation of deep or open sections of the flight path omitted from analysis due to fog. b, Study thaw bulbs beneath old lakes, large rivers and high-discharge regions, yedoma deposits and hydrocarbon basins. Further map source springs substantially increases permafrost permeability locally27,28. information is provided in the Supplementary Information. Few additional For this region we reasoned H1: sites with subcap seeps should records of natural fossil methane seepage in the terrestrial Arctic have been be disproportionately associated with low-ground ice content and documented in Canada and Russia6,16. fluvial deposits, substrate characteristics most likely to enhance thaw-bulb formation and permeability. cryosphere cap would lead to transient release of methane trapped Region 2 (southcentral Alaska, Fig. 2b) was covered by glaciers beneath and within permafrost or through faults, joints and and ice sheets during the Last Glacial Maximum29, but today has fractures previously sealed by ice, hydrates or the normal stress wasting glaciers and discontinuous, sporadic or no permafrost30.

ARTICLES NATURE GEOSCIENCE DOI: 10.1038/NGEO1480

a Subcap 60 Superficial (yedoma-type) 40 Superficial (non-yedoma) surveyed 2 20 km Number of subcap

seep sites per 1,000 0 b 3,000

2,000

surveyed 1,000 Number of lakes 0 c 10 ) survey 8 Geos eeps ¬1 ¬2 6 yedom a km

4 non-y edom a area d 4

Seep emissions 2 (kg CH 0

d 3,000 Alaska and Alaska North Slope Brooks Range Alaska interior 2,000 Chugach ranges Yedoma-type Thermokarst Colville R. CH4 thermokarst lakes Lakes and 1,000 lakes CH4 Yukon R. fjords sea level (m) CH CH CH Elevation above 4 4 4 CH4 CH4 CH4 0 10 100 1,000 surface (m)

Depth below 10,000

72° N Distance along a North¬South transect 60° N

Permafrost Yedoma-type Fault Coal Bedrock Conventional Glacial Glacier deposit seam cap hydrocarbon buried Hydrate stability zone organics Top Bottom

Figure 4 | Subcap (macroseep) and superficial methane seep emissions in Alaska. a–d, In 50-km-wide bins along the north–south Alaska transect: the number of subcap-seep sites normalized by survey area (a); total number of lakes surveyed (b); methane emissions from subcap macroseeps and superficial seeps (yedoma-type and non-yedoma lakes) (c); and a schematic cross-section (not to scale) of major topographic and subsurface features (d). Features include permafrost distribution and thickness, examples of open and closed thaw bulbs in conjunction with faults and permeable strata providing gas migration conduits, examples of primary structural and stratigraphic gas traps beneath the secondary cryosphere cap, and potential methane sources contributing to ebullition emissions in Alaska including superficial ecosystem methane from surface lake and taberal (thawed permafrost) sediments and five subcap sources: microbial methane in buried glacial sediments, coalbed methane (biogenic or thermogenic), thermogenic methane from deep conventional hydrocarbons, and possibly methane derived from hydrate dissociation.

As differential ice loading can overpressurize fluid reservoirs and Finally, we reasoned that given equal hydrocarbon seepage from cause sediment fracturing beneath ice sheets11, and because the underlying geologic structures, the magnitude and duration of loss of glacial ice reduces normal stress on ground, opens joints subcap gas seepage is a function of cap longevity. Long-lived caps and activates faults and fissures, thereby increasing permeability of (for example, thick, continuous permafrost and continental ice the crust to fluid flow10,14,31,32, we reasoned H2: in the previously caps) trap more gas, leading to larger seep emissions when conduits glaciated region of southcentral Alaska, where glacial wastage open when compared with short-lived permafrost and glaciers that continues at present, subcap seeps should be disproportionately form and disintegrate on glacial–interglacial or shorter timescales. associated with neotectonic faults. H4: geologic seeps in the continuous permafrost region should be Region 3 (interior Alaska, Fig. 2b) is characterized by a larger than seeps in the southcentral, previously glaciated region cryosphere cap that degraded from thinner, but continuous of Alaska; and, subcap seep sites in southcentral Alaska should be permafrost during the Late Pleistocene glacial period to a strongly disproportionately proximate to modern glaciers. disintegrated discontinuous to isolated permafrost today30. H3: in Geospatial analysis confirmed hypothesis H1. In the ice-rich interior Alaska, where permafrost disintegration is advanced to continuous permafrost zone of northern Alaska, subcap seep sites various levels of discontinuity for centuries to millennia, geologic were disproportionately located in isolated areas of low-ice content seeps should be relatively rare. They are more likely to occur and fluvial deposits beneath deep lakes or near rivers (Fig. 5a and in association with neotectonic activity unrelated to cryosphere Supplementary Table S1), features that together suggest steep thaw dynamics, with site-specific geological controls similar to those of gradients. Examples include numerous seep fields identified in lakes regions outside the Arctic8,10,21,24–26. in the vicinities of known coal and natural gas deposits in northern

NATURE GEOSCIENCE DOI: 10.1038/NGEO1480 ARTICLES

acNorthern continuous permafrost Southcentral glaciated Alaska eSouthcentral glaciated Alaska 10 30 12 8 20 6 8 4 10 4 2 Number of seep sites Number of seep sites Number of seep sites 0 0 0 0102030405060 0102030405060 0102030405060 Nearest fluvial deposit (km) Nearest fault (km) Nearest glacier (km)

bd120 120 f300

90 80 200 60 40 100 30 Number of lakes Number of lakes Number of lakes

0 0 0 0102030405060 0102030405060 0102030405060 Nearest fluvial deposit (km) Nearest fault (km) Nearest glacier (km)

Figure 5 | Spatial association of subcap-seep sites with fluvial deposits in northern continuous permafrost and with faults near glaciers in southcentral Alaska. a,b, The 30 identified subcap-seep sites in the northern Alaska continuous permafrost zone (a) were disproportionately close to fluvial deposits compared with a random selection of 1,000 lakes in this region (b)(p < 0.001). c,d, The 63 identified subcap-seep sites in southcentral Alaska had a stronger association with faults (c) than 1,000 randomly selected lakes in this zone (d)(p < 0.001). e,f, Southcentral Alaska subcap-seep sites were disproportionately close to the boundaries of modern glaciers (e) compared with a random selection of 1,000 lakes (f)(p < 0.001). See Supplementary Figs S3–S5.

Alaska4,33 adjacent to the Meade River, in the drainages of the in nearby deep gas wells (Fig. 3c,d) and in hydrocarbon reservoirs Brooks Range, and along the upper Colville River where recent common throughout Alaska33,36. In two instances, we found geophysical investigation determined a deep or through-going thaw detectable levels of 14C in high-discharge subcap seeps: Lake Eyak, bulb (Kanevskiy M., unpublished data, written communication; in the previously glaciated zone of southcentral Alaska, and lakes in July 20, 2011). The relationship of subcap seeps to fluvial deposits Greenland that were covered by glacial ice during the LIA. In Lake was stronger in the northern continuous permafrost region than Eyak (Fig. 6) and across Alaska (Supplementary Fig. S1), we found a in the rest of Alaska (Supplementary Table S2), possibly because relationship between methane stable isotopes, radiocarbon age and there are other potential escape routes for gas, such as surface distance to faults. Faults seem to allow the escape of deeper, more faults and unfrozen, unconsolidated sediments in the interior and 14C-depleted methane to the atmosphere, whereas seeps away from southcentral regions. faults entrained 14C-enriched methane formed in shallower sedi- In southcentral Alaska, where post-Little Ice Age (LIA; about ments from microbial decomposition of younger organic matter. ad 1650–1850) glacial wastage and associated isostatic rebound In Greenland, highly ebullitive (35–847 l d−1) subcap seeps were enhance faulting and seismicity31,32, our analysis confirmed H2. confined to lakes within a 0–2 km belt along the present ice-sheet Subcap seep sites were associated with faults within a 7 km belt from boundary (Supplementary Fig. S2). The zone of observed seeps was the modern glacial extent (Fig. 5c–f). Most seeps were located in exactly co-extensive with the zone of ice-sheet retreat over the past areas affected by seismicity from isostatic rebound associated with 100 years following its LIA advance37. The geographic locations, 13 deglaciation following the LIA (refs 34,35). stable isotope values (δ CCH4, −66.8 to −62.7%; δDCH4, −402 to The diminishing frequency of subcap seeps with distance −359%), lack of ethane and radiocarbon ages of Greenland subcap from modern glaciation (H4), their paucity in interior Alaska seeps (1,420–1,530 yr bp) suggest that methane probably formed by (H3) except along the boundary of continuous–discontinuous microbial decomposition of organic matter buried and capped by permafrost (Fig. 4), and the exceedingly higher subcap seepage the LIA and previous neoglacial advances. Retreat of the ice sheet rates observed in the northern Alaska continuous permafrost zone since the LIA allows for the present escape of this relatively younger (H4, mean ± s.e.m., n sites; minimum to maximum: 325 ± subcap methane to the atmosphere. −1 −1 −1 −1 260 kg CH4 site d , n = 30 sites; 0.3 to 7,845 kg CH4 site d ; First-order extrapolation of geospatial seep relationships −1 Fig. 2) when compared with the previously glaciated southcentral across Alaska for superficial seeps (0.75 ± 0.12 Tg CH4 yr ), −1 −1 −1 Alaska zone (12 ± 2 kg CH4 site d , n = 40 sites; 0.2 to 57 kg subcap macroseeps (0.08 ± 0.01 Tg CH4 yr ) and miniseeps −1 −1 −1 CH4 site d ) imply a transient duration for seepage associated (0.17±0.03 Tg CH4 yr ), and other potential non-subcap geologic with cryosphere degradation (Supplementary Information). microseepage from unfrozen soils overlying hydrocarbon-prone −1 sedimentary basins (0.5–1.1 Tg CH4 yr ) after (25), results in 14 −1 C-depleted methane release from seeps to the atmosphere an estimate of 1.5–2 Tg CH4 yr (Supplementary Table S3). Both subcap seeps and some superficial seeps (for example, those Our extrapolation increases the present estimate of Alaska’s 14 −1 in yedoma-type lakes) release C-depleted methane to the atmo- natural methane emissions (3 Tg CH4 yr ; ref. 38) to the sphere (Fig. 3b). Whereas the isotope geochemistry of all superficial atmosphere by 50–70%. seeps showed clear microbial origin (Fig. 3c), that of the subcap A conservative first-order extrapolation based on geospatial re- seeps varied. Most subcap seeps emitted 14C-dead methane. Their lationships observed in the northern Alaska continuous permafrost stable isotope and geochemical signatures were consistent with region, following scaling rules of refs 26 and 39 (Supplementary microbially produced coal bed or thermogenic methane observed Information) would place the magnitude of subcap macroseep

ARTICLES NATURE GEOSCIENCE DOI: 10.1038/NGEO1480

ab¬80 ¬80

¬70 ¬70

¬60 ¬60 () () 4 ¬50 4 ¬50 CH CH C C 13 13

δ δ ¬40 δ 13 ¬40 CCH4() ¬75 to ¬65 ¬30 ¬65 to ¬55 ¬30 ¬55 to ¬45 ¬45 to ¬35 ¬20 ¬20 0204060 ¬280 ¬230 ¬180 ¬130 pmC (%) δ DCH4()

2 km 4 km

Figure 6 | Subcap seep methane stable isotopes and radiocarbon age in relation to faults in the Lake Eyak region of southcentral Alaska. a–d, The 13 14 2 relationship between subcap seep δ CCH4 and C content (pmC, r = 0.91) (a) and δDCH4 (b), and proximity to known faults near Lake Eyak (c) and in lakes extending east 100 km to Katalla (d), where oil and gas were first discovered in Alaska in 1902, suggests that deep, 14C-dead, thermogenic methane escapes through faults, whereas gas seeping through smaller fractures and fissures mixes with microbial methane in shallower, 14C-enriched organics. The neotectonic fault data (d) were provided by T. L. Pavlis (personal communication, May 14, 2010 and ref. 48), but not all faults and surface ruptures are known. Black lines are mapped faults; white dots are unsampled, mapped seep fields.

−1 −1 (0.7 ± 0.1 Tg CH4 yr ) and miniseep (1.4 ± 0.3 Tg CH4 yr ) made to this hypothesis that in a warmer world, thawing permafrost emissions from lakes in the natural gas-rich region of the terrestrial and wastage of glaciers and ice sheets could lead to a significant 2 ◦ −1 Arctic north of 60 N at ∼2 Tg CH4 yr . Emissions associated with transitional degassing of subcap methane. Deep permafrost thaw glacier and ice-sheet wastage are not included in this estimate, but is projected to occur over centuries to millennia1,40. However, are a further source of atmospheric methane in the pan-Arctic. already ongoing permafrost warming41 leading to higher unfrozen In glaciated regions, cryosphere advances and retreats increase water content, intensification of surface water and subpermafrost fracturing of sedimentary rocks10,11,31,32. Where these rocks host groundwater exchange42, expansion of existing taliks and formation natural gas accumulations, this opening of new and pre-existing of new taliks may increase permeability to gas flow locally on lithospheric conduits triggers the migration of gas to the surface, shorter timescales (Supplementary Information). It is possible thus forming 14C-depleted gas seeps when ice retreats. In permafrost that methane oxidation processes could temper future subcap regions, disintegration of permafrost removes the ice seal from emissions, but at the very least, an injection of methane carbon due pre-existing faults and from unconsolidated sediments, also to cryosphere degradation will increase surface carbon cycling. creating conduits through the lithosphere that allow gas previously We have shown that cryosphere-cap methane seeps are prevalent trapped by permafrost to escape to the atmosphere. Accordingly, in high-latitude terrestrial environments along steep thaw gradients pan-Arctic regions are a special place for gas seepage due to past in permafrost and in association with wasting glaciers, often and ongoing cryosphere disintegration. This may explain why our through activated faults. The geospatial association of these seeps −1 subcap seep emission estimate (∼2 Tg CH4 yr ) is relatively large in with cryosphere boundaries in Alaska and Greenland suggests comparison with the global geologic seep emission estimates (∼3–4 that if this relationship holds true for other regions where −1 Tg CH4 yr ; ref. 26). sedimentary basins are at present capped by permafrost, glaciers As our geospatial and geochemical field data support the hypoth- and ice sheets, such as northern West Siberia, rich in natural esis that cryosphere degradation leads to the release of 14C-depleted gas and partially underlain by thin permafrost predicted to methane previously trapped by the cryosphere, an extension can be degrade substantially by 2100 (ref. 40), a very strong increase in

NATURE GEOSCIENCE DOI: 10.1038/NGEO1480 ARTICLES methane carbon cycling will result, with potential implications for 2. Gautier, D. L. et al. Assessment of undiscovered oil and gas in the arctic. Science climate warming feedbacks. 324, 1174–1179 (2009). 3. Collett, T. S. et al. Permafrost-associated natural gas hydrate occurrences on Methods the Alaska North Slope. Mar. Petrol. Geol. 28, 279–294 (2011). 4. Flores, R. M., Stricker, G. D. & Kinney, S. A. Alaska Coal Geology, Resources, Aerial surveys as an approach to subcap seep detection and mapping. We 2 and Coalbed Methane Potential (US Dept. Interior Rep., USGS Digital Data conducted aerial surveys on approximately 6,700 lakes in an 11,260 km area along Series DDS-77, v.1.0., 2004). a north–south transect in Alaska during ice-cover seasons from 2008 to 2010 5. Isaksen, I. S. A., Gauss, M., Myhre, G., Walter Anthony, K. M. & Ruppel, C. (Fig. 2). From the aeroplane directly and from photographs obtained during flights, Strong atmospheric chemistry feedback to climate warming from Arctic we counted the number of open holes observed in the ice cover of frozen water methane emissions. Glob. Biogeochem. Cycles 25, GB2002 (2011). bodies (usually lakes). We classified the likelihood of open ice holes indicating 6. Clarke, R. & Cleverly, R. in Petroleum Migration (eds England, W. & subcap seeps as ‘very likely’ or ‘maybe’ at each site according to the number and Fleet, A.) 265–271 (Geological Society Special Publication No. 59, Geological size of holes observed, their morphology, clustering and distribution, and the Society, 1991). overall ice conditions on lakes. Ground-truth expeditions were conducted in 7. Hunt, J. M. Petroleum Geochemistry and Geology 2nd edn concert with aerial-survey flights (Supplementary Table S4). Of the 290 sites for (W.H. Freeman, 1996). which we initially observed seep-size open water holes, 77 sites were classified as 8. Lacroix, A. V. Unaccounted for sources of fossil and isotopically-enriched ‘very likely’ containing subcap seeps and were the basis of our emission estimates. methane and their contribution to the emissions inventory. Chemosphere 26, We ground-truthed 50 of the 77 ‘very likely’ sites and confirmed that subcap seeps 507–557 (1993). were present at all of these sites on the basis of observations of high ebullition 9. Romanovskii, N. N. et al. Environmental evolution in the Laptev Sea region (>10 L seep−1 d−1 on 72 measured seeps) and isotopic and geochemical gas during Late Pleistocene and Holocene. Polarforschung 68, 237–245 (2000). composition (Fig. 3). 10. Etiope, G., Milkov, A. & Derbyshire, E. Did geologic emissions of methane play Estimating subcap seep emissions in Alaska and the pan-Arctic. At individual any role in Quaternary climate change? Glob. Planet. Change 22, 79–88 (2008). subcap seep sites we counted the number of bubble streams (seeps) per open hole 11. Lerche, I., Yu, Z., Torudbakken, B. & Thomsen, R. O. Ice loading effects in and measured fluxes in replicated holes to statistically estimate the number of sedimentary basins with reference to the Barents Sea. Mar. Petrol. Geol. 14, ‘macroseeps’ and their cumulative flux per site. For sites where we counted open 277–338 (1997). holes but did not measure flux, we applied the mean fluxes measured at nearby sites. 12. US Environmental Protection Agency. Methane and Nitrous Oxide Emissions To analyse subcap macroseep emissions by latitude, we divided our Alaska From Natural Sources (US EPA, Office of Atmospheric Programs, Climate north–south flight survey transect into 25 latitudinal bins (50 km in north–south Change Division, 2010). direction and 1 km left and right of the flight transect) and used the intersect 13. Formolo, M. J., Salacup, J. M., Petsch, S. T., Martini, A. M. & Nüsslein, K. tool in Esri ArcGIS 9.3 to determine the total land surface area, lake area and A new model linking atmospheric methane sources to Pleistocene glaciation number of lakes surveyed within each bin43. Following the European Monitoring via methanogenesis in sedimentary basins. Geology 36, 139–142 (2008). and Evaluation Programme/European Environment Agency guidelines44, we 14. Grassmann, S. et al. pT-effects of Pleistocene glacial periods on permafrost, converted our point-source emissions into an areal extent by aggregating the gas hydrate stability zones and reservoir of the Mittelplate oil field, northern observed point-source emissions within each latitudinal bin and dividing by the Germany. Mar. Petrol. Geol. 27, 298–306 (2010). area surveyed within each bin (Fig. 4). 15. Yakushev, V. S. & Chuvilin, E. M. Natural gas and hydrate accumulations We derived a first-order estimate of subcap macroseep emissions in Alaska within permafrost in Russia. Cold Regions Sci. Technol. 31, 189–197 (2000). −1 (0.08±0.1 Tg CH4 yr ) following the scaling guidelines for point-source emissions 16. Bowen, R. G., Dallimore, S. R., Cote, M. M., Wright, J. F. & Lorenson, T. D. of ref. 44 and ref. 39. We assumed that the point-source macroseeps observed in Proc. Ninth International Conference on Permafrost (eds Kane, D. L. & along the flight surveys in each of the northern Alaska continuous permafrost Hinkel, K. M.) 171–176 (Institute of Northern Engineering, 2008). −1 2 region (3,562,800 kg CH4 yr for 793 km of lakes surveyed), the interior Alaska 17. Walter, K. M., Zimov, S. A., Chanton, J. P., Verbyla, D. & Chapin, F. S. III. −1 2 region (6,951 kg CH4 yr for 38 km of lakes surveyed) and the previously Methane bubbling from Siberian thaw lakes as a positive feedback to climate −1 2 glaciated southcentral Alaska region (168,931 kg CH4 yr for 169 km of lakes warming. Nature 443, 71–75 (2006). surveyed) were a representative sample of subcap seeps for all lakes in these zones 18. Bastviken, D. L., Tranvik, J., Downing, J. A., Crill, P. M. & Enrich-Prast, A. across the rest of Alaska. We quantified the uncertainty of the emission value by Freshwater methane emissions offset the continental carbon sink. Science propagating errors of the estimates of number of seeps per site and seep flux (kg 331, 50 (2011). −1 −1 CH4 seep d ) for all measured and estimated sites in the three Alaska study 19. Zimov, S. A. et al. North Siberian lakes: A methane source fueled by Pleistocene regions at the 95% confidence limits. carbon. Science 277, 800–802 (1997). In addition, we observed 103–104 ebullition ‘miniseeps’ at each of eight subcap 20. Walter Anthony, K. M. et al. Estimating methane emissions from northern sites visited in summer. Miniseeps were isotopically similar to macroseeps, but lakes using ice bubble surveys. Limnol. Oceanogr. Methods 8, 592–609 (2010). were invisible in winter owing to lower bubbling rates. We quantified miniseepage 21. Judd, A. G. Natural seabed gas seeps as sources of atmospheric methane. by measuring miniseep densities and ebullition rates at three sites. To scale up Environ. Geol. 46, 988–996 (2004). subcap emissions to the state of Alaska, we assumed that the observed ratio 22. Etiope, G. Natural emissions of methane from geological seepage in Europe. of miniseep-to-macroseep fluxes (2.0 ± 0.4, standard deviation) applies to all Atm. Environ. 43, 1430–1443 (2009). Alaska macroseep sites. 23. Reeburgh, W. S. Oceanic methane biogeochemistry. Chem. Rev. 107, Following the scaling methods of refs 39 and 44, we estimated the potential 486–513 (2007). magnitude of present-day subcap methane emissions from pan-Arctic lakes 24. Etiope, G. & Klusman, R. W. Geologic emissions of methane to the atmosphere. overlying assumed natural gas fields in the continuous permafrost zone north Chemosphere 49, 777–789 (2002). ◦ of ∼60 N on the basis of the US Geological Survey’s Circum-Arctic Resource 25. Etiope, G. & Klusman, R. Microseepage in drylands: Flux and implications in 2 −1 Appraisal . Our first-order estimate of 0.7±0.1 Tg CH4 yr for macroseep subcap the global atmospheric source/sink budget of methane. Glob. Planet. Change emissions is the product of the emission factor observed along our survey flight 72, 265–274 (2010). path in the northern Alaska continuous permafrost zone and the area of large 26. Etiope, G., Lassey, K. R., Klusman, R. & Boschi, E. Re-appraisal of the fossil pan-Arctic lakes overlying assumed natural gas fields2,45 (150,000 km2). Assuming methane budget and related emission from geologic sources. Geophys. Res. Lett. miniseep emissions proportionate to the Alaska observations, total subcap 35, L09307 (2008). emissions in the northern continuous permafrost region of the pan-Arctic would 27. Johnston, G. H. & Brown, R. G. B. Some observations on permafrost −1 be 2±0.3 Tg CH4 yr . Further subcap emissions in the pan-Arctic associated with distribution at a lake in the MacKenzie Delta, N.W.T., Canada. Arctic 17, ◦ permafrost degradation south of 60 N and with wastage of glaciers and ice sheets 162–175 (1964). were not assessed, but would further increase the pan-Arctic emission estimate. 28. Yoshikawa, K., Hinzman, L. D. & Kane, D. L. Spring and aufeis (icing) Geospatial and other analyses. We used the one-sided, two-sample hydrology in Brooks Range, Alaska. J. Geophys. Res. 112, G04S43 (2007). Kolmogorov–Smirnov test for geospatial hypothesis testing (Fig. 5 and 29. Dyke, A. S., Moore, A. & Robertson, L. Deglaciation of North America Supplementary Figs S3–S5). The full set of geospatial analyses and details of (Geological Survey of Canada Open File 1574, 2003). laboratory procedures, field work, scaling methods and statistics for subcap and 30. Jorgenson, T. et al. Proc. Ninth International Conference on Permafrost superficial seeps are provided in Supplementary Methods. (eds. Kane, D. L. & Hinkel, K. M.) map in scale 1:7,000,000 (Institute of Northern Engineering, Fairbanks, 2008). Received 1 August 2011; accepted 19 April 2012; published online 31. Sauber, J. M. & Molnia, B. F. Glacier ice mass fluctuations and fault instability 20 May 2012 in tectonically active Southern Alaska. Glob. Planet. Change 42, 279–293 (2004). 32. Elliott, J. L., Larsen, C. F., Freymueller, J. T. & Motyka, R. J. Tectonic References block motion and glacial isostatic adjustment in southeast Alaska and 1. McGuire, D. A. et al. Sensitivity of the carbon cycle in the Arctic to climate adjacent Canada constrained by GPS measurements. J. Geophys. Res. 115, change. Ecol. Monogr. 79, 523–555 (2009). B09407 (2010).

ARTICLES NATURE GEOSCIENCE DOI: 10.1038/NGEO1480

33. Burruss, R. C., Lillis, P. G. & Collett, T. S. Geochemistry of Natural Gas, North Communication (comp.) 2003, Circumpolar Active-Layer Permafrost System, Slope, Alaska: Implications for Future Oil and Gas Resources, NPRA (US Dept Version 2.0 (eds Parsons, M. & Zhang, T.) (National Snow and Ice Data Interior Rep., USGS Open-File Report 03-041, 2003). Center/World Data Center for Glaciology, 1998). 34. Larsen, C. F., Motyka, R. J., Freymueller, J. T., Echelmeyer, K. A. & Ivins, E. R. 46. Etiope, G., Baciu, C. L. & Schoell, M. Extreme methane deuterium, nitrogen, Rapid viscoelastic uplift in southeast Alaska caused by post-Little Ice Age glacial and helium enrichment in natural gas from the Homorod seep (Romania). retreat. Earth Planet. Sci. Lett. 237, 548–560 (2005). Chem. Geol. 280, 89–96 (2011). 35. Molnia, B. F. Late nineteenth to early twenty-first century behavior of Alaskan 47. Milkov, A. V. Worldwide distribution and significance of secondary microbial glaciers as indicators of changing regional climate. Glob. Planet. Change 56, methane formed during petroleum biodegradation in conventional reservoirs. 23–56 (2007). Org. Geochem. 42, 184–207 (2011). 36. Claypool, G. E., Threlkeld, C. N. & Magoon, L. B. Biogenic and thermogenic 48. Pavlis, T. L. & Bruhn, R. L. Application of LIDAR to resolving bedrock origins of natural gas in Cook Inlet Basin, Alaska. AAPG Bull. 64, structure in areas of poor exposure: An example from the STEEP study area, 1131–1139 (1980). southern Alaska. Geol. Soc. Am. Bull. 123, 206–217 (2011). 37. Forman, S. L., Marin, L., Van der Veen, C., Tremper, C. & Csatho, B. 49. Kampman, N. et al. Pulses of carbon dioxide emissions from intracrustal faults Little Ice Age and neoglacial landforms at the Inland Ice margin, Isungguata following climatic warming. Nature Geosci. 5, 352–358 (2012). Sermia, Kangerlussuaq, west Greenland. Boreas 36, 341–351 (2007). 38. Zhuang, Q. et al. Net emissions of CH4 and CO2 in Alaska: Implications for the Acknowledgements region’s greenhouse gas budget. Ecol. Appl. 17, 203–212 (2007). We thank researchers at the Alaska DGGS and the USGS for contributions to data sets; 39. Etiope, G., Fridriksson, T., Italiano, F., Winwarter, W. & Theloke, J. Natural D. Whiteman, L. McFadden and A. Strohm for field assistance; L. Oxtoby, C. Langford emissions of methane from geothermal and volcanic sources in Europe. and D. Fields for laboratory work. V. Romanovsky, F. S. Chapin III, T. Pavlis and G. J. Volcan. Geoth. Res. 165, 76–86 (2007). Etiope provided valuable comments on the manuscript. This work was supported by DOE 40. Romanovsky, V. E. et al. in Proc. of the Ninth International Conference on #DE-NT0005665, NASA Carbon Cycle Sciences, the NASA Astrobiology Institute’s Icy Permafrost (eds Kane, D. L. & Hinkel, K. M.) 1511–1518 (Institute of Northern Worlds node, the NSF Division of Earth Sciences and the NSF Arctic Division. Engineering, 2008). 41. Romanovsky, V. E. et al. Thermal state of permafrost in Russia. Permafrost Periglac. Process 21, 136–155 (2010). Author contributions 42. Rowland, J. C., Travis, B. J. & Wilson, C. J. The role of advective heat transport K.M.W.A. wrote the paper. K.M.W.A. and P.A. designed the experiment, conducted the in talik development beneath lakes and ponds in discontinuous permafrost. field work and performed the seep analyses. G.G. provided expertise on cryosphere pro- Geophys. Res. Lett. 38, L17504 (2011). cesses. Isotopic analysis was conducted in the laboratory of J.C. All authors commented on 43. Arp, C. D. & Jones, B. M. Geography of Alaska Lake Districts: Identification, the analysis, interpretation and presentation of the data, and were involved in the writing. Description, and Analysis of Lake-Rich Regions of a Diverse and Dynamic State (US Dept Interior, USGS Scientific Investigations Report 40, 2009). Additional information 44. European Environment Agency. EMEP/EEA Air Pollutant Emission Inventory The authors declare no competing financial interests. Supplementary information Guidebook. EEA Technical Report (2009). accompanies this paper on www.nature.com/naturegeoscience. Reprints and permissions 45. Brown, J., Ferrians, O. J. Jr, Heginbottom, J. A. & Melnikov, E. in International information is available online at www.nature.com/reprints. Correspondence and Permafrost Association Standing Committee on Data Information and requests for materials should be addressed to K.M.W.A.